Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -10,112 +10,175 @@ from urllib.parse import quote
|
|
10 |
from openai import OpenAI
|
11 |
|
12 |
# ------------------ App Configuration ------------------
|
13 |
-
st.set_page_config(page_title="Schlaeger Forrestdale
|
14 |
st.title("π Schlaeger Forrestdale Document Assistant")
|
15 |
-
st.caption("Explore City of Armadale construction documents using AI + OCR
|
16 |
|
17 |
-
# ------------------ Load API Key
|
18 |
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
|
19 |
-
|
20 |
-
|
21 |
-
if not OPENAI_API_KEY or not ASSISTANT_ID:
|
22 |
-
st.error("β Missing secrets. Please set both OPENAI_API_KEY and ASSISTANT_ID in Hugging Face Space secrets.")
|
23 |
st.stop()
|
24 |
|
25 |
client = OpenAI(api_key=OPENAI_API_KEY)
|
26 |
|
27 |
-
# ------------------
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
if "
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
st.
|
39 |
-
|
40 |
-
|
41 |
-
st.
|
42 |
-
st.
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
|
74 |
-
|
75 |
-
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
|
94 |
-
|
95 |
-
|
96 |
-
|
97 |
-
# ------------------ Sorting and Filters ------------------
|
98 |
-
if st.session_state.results:
|
99 |
-
disciplines = sorted(set(d.get("discipline", "") for d in st.session_state.results))
|
100 |
-
selected_discipline = st.selectbox("π Filter by discipline", ["All"] + disciplines)
|
101 |
-
page_size = 8
|
102 |
-
page_num = st.number_input("Page", min_value=1, step=1, value=1)
|
103 |
-
|
104 |
-
filtered_results = [r for r in st.session_state.results if selected_discipline == "All" or r.get("discipline") == selected_discipline]
|
105 |
-
paged = filtered_results[(page_num - 1) * page_size : page_num * page_size]
|
106 |
-
|
107 |
-
st.markdown("---")
|
108 |
-
st.subheader("π Drawing Results")
|
109 |
-
cols = st.columns(4)
|
110 |
-
|
111 |
-
for i, item in enumerate(paged):
|
112 |
-
with cols[i % 4]:
|
113 |
-
st.markdown(f"**{item['drawing_number']}**")
|
114 |
-
st.markdown(f"_Discipline: {item['discipline']}_")
|
115 |
-
st.caption(item.get("summary", ""))
|
116 |
-
for url in item.get("images", [])[:1]:
|
117 |
-
st.image(url, caption=f"{item['drawing_number']} β Page 1", use_column_width=True)
|
118 |
-
else:
|
119 |
-
for msg in st.session_state.messages:
|
120 |
with st.chat_message(msg["role"]):
|
121 |
st.markdown(msg["content"], unsafe_allow_html=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
from openai import OpenAI
|
11 |
|
12 |
# ------------------ App Configuration ------------------
|
13 |
+
st.set_page_config(page_title="Schlaeger Forrestdale DocAIA", layout="wide", initial_sidebar_state="collapsed")
|
14 |
st.title("π Schlaeger Forrestdale Document Assistant")
|
15 |
+
st.caption("Explore City of Armadale construction documents using AI + OCR π§ ")
|
16 |
|
17 |
+
# ------------------ Load API Key ------------------
|
18 |
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
|
19 |
+
if not OPENAI_API_KEY:
|
20 |
+
st.error("β Missing OPENAI_API_KEY in environment variables.")
|
|
|
|
|
21 |
st.stop()
|
22 |
|
23 |
client = OpenAI(api_key=OPENAI_API_KEY)
|
24 |
|
25 |
+
# ------------------ Tabs ------------------
|
26 |
+
tab1, tab2 = st.tabs(["π Contract", "π Technical"])
|
27 |
+
|
28 |
+
# ------------------ Contract Assistant ------------------
|
29 |
+
with tab1:
|
30 |
+
ASSISTANT_ID = "asst_KsQRedoJUnEeStzfox1o06lO"
|
31 |
+
|
32 |
+
if "contract_messages" not in st.session_state:
|
33 |
+
st.session_state.contract_messages = []
|
34 |
+
if "contract_thread_id" not in st.session_state:
|
35 |
+
st.session_state.contract_thread_id = None
|
36 |
+
if "contract_image_url" not in st.session_state:
|
37 |
+
st.session_state.contract_image_url = None
|
38 |
+
|
39 |
+
st.markdown("### π§ Ask a Contract-Specific Question")
|
40 |
+
contract_prompt = st.chat_input("Example: What is the defects liability period?", key="contract_input")
|
41 |
+
|
42 |
+
if contract_prompt:
|
43 |
+
st.session_state.contract_messages.append({"role": "user", "content": contract_prompt})
|
44 |
+
|
45 |
+
if st.session_state.contract_messages and st.session_state.contract_messages[-1]["role"] == "user":
|
46 |
+
try:
|
47 |
+
if st.session_state.contract_thread_id is None:
|
48 |
+
thread = client.beta.threads.create()
|
49 |
+
st.session_state.contract_thread_id = thread.id
|
50 |
+
|
51 |
+
client.beta.threads.messages.create(
|
52 |
+
thread_id=st.session_state.contract_thread_id,
|
53 |
+
role="user",
|
54 |
+
content=st.session_state.contract_messages[-1]["content"]
|
55 |
+
)
|
56 |
+
|
57 |
+
run = client.beta.threads.runs.create(
|
58 |
+
thread_id=st.session_state.contract_thread_id,
|
59 |
+
assistant_id=ASSISTANT_ID
|
60 |
+
)
|
61 |
+
|
62 |
+
with st.spinner("π€ Parsing contract documents..."):
|
63 |
+
while True:
|
64 |
+
run_status = client.beta.threads.runs.retrieve(
|
65 |
+
thread_id=st.session_state.contract_thread_id,
|
66 |
+
run_id=run.id
|
67 |
+
)
|
68 |
+
if run_status.status in ("completed", "failed", "cancelled"):
|
69 |
+
break
|
70 |
+
time.sleep(1)
|
71 |
+
|
72 |
+
if run_status.status == "completed":
|
73 |
+
messages = client.beta.threads.messages.list(thread_id=st.session_state.contract_thread_id)
|
74 |
+
for message in reversed(messages.data):
|
75 |
+
if message.role == "assistant":
|
76 |
+
reply = message.content[0].text.value
|
77 |
+
st.session_state.contract_messages.append({"role": "assistant", "content": reply})
|
78 |
+
|
79 |
+
match = re.search(r'Document Reference:\s*(.*?),\s*Page\s*(\\d+)', reply)
|
80 |
+
if match:
|
81 |
+
doc_name = match.group(1).strip()
|
82 |
+
page = int(match.group(2))
|
83 |
+
page_str = f"{page:04d}"
|
84 |
+
folder = quote(doc_name)
|
85 |
+
url = f"https://raw.githubusercontent.com/AndrewLORTech/c2ozschlaegerforrestdale/main/{folder}/{folder}_page_{page_str}.png"
|
86 |
+
st.session_state.contract_image_url = url
|
87 |
+
break
|
88 |
+
else:
|
89 |
+
st.error(f"β Contract Assistant failed: {run_status.status}")
|
90 |
+
st.rerun()
|
91 |
+
except Exception as e:
|
92 |
+
st.error(f"β Error: {e}")
|
93 |
+
|
94 |
+
for msg in st.session_state.contract_messages:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
95 |
with st.chat_message(msg["role"]):
|
96 |
st.markdown(msg["content"], unsafe_allow_html=True)
|
97 |
+
|
98 |
+
if st.session_state.contract_image_url:
|
99 |
+
st.image(st.session_state.contract_image_url, caption="π Contract Page Image", use_container_width=True)
|
100 |
+
|
101 |
+
# ------------------ Technical Drawing Assistant ------------------
|
102 |
+
with tab2:
|
103 |
+
ASSISTANT_ID = "asst_DjvuWBc7tCvMbAhY7n1em4BZ"
|
104 |
+
|
105 |
+
if "tech_messages" not in st.session_state:
|
106 |
+
st.session_state.tech_messages = []
|
107 |
+
if "tech_thread_id" not in st.session_state:
|
108 |
+
st.session_state.tech_thread_id = None
|
109 |
+
if "tech_results" not in st.session_state:
|
110 |
+
st.session_state.tech_results = []
|
111 |
+
|
112 |
+
tech_prompt = st.chat_input("Ask about plans, drawings or components", key="tech_input")
|
113 |
+
if tech_prompt:
|
114 |
+
st.session_state.tech_messages.append({"role": "user", "content": tech_prompt})
|
115 |
+
|
116 |
+
if st.session_state.tech_messages and st.session_state.tech_messages[-1]["role"] == "user":
|
117 |
+
try:
|
118 |
+
if st.session_state.tech_thread_id is None:
|
119 |
+
thread = client.beta.threads.create()
|
120 |
+
st.session_state.tech_thread_id = thread.id
|
121 |
+
|
122 |
+
client.beta.threads.messages.create(
|
123 |
+
thread_id=st.session_state.tech_thread_id,
|
124 |
+
role="user",
|
125 |
+
content=st.session_state.tech_messages[-1]["content"]
|
126 |
+
)
|
127 |
+
|
128 |
+
run = client.beta.threads.runs.create(
|
129 |
+
thread_id=st.session_state.tech_thread_id,
|
130 |
+
assistant_id=ASSISTANT_ID
|
131 |
+
)
|
132 |
+
|
133 |
+
with st.spinner("π€ Searching technical drawings..."):
|
134 |
+
while True:
|
135 |
+
run_status = client.beta.threads.runs.retrieve(
|
136 |
+
thread_id=st.session_state.tech_thread_id,
|
137 |
+
run_id=run.id
|
138 |
+
)
|
139 |
+
if run_status.status in ("completed", "failed", "cancelled"):
|
140 |
+
break
|
141 |
+
time.sleep(1)
|
142 |
+
|
143 |
+
if run_status.status == "completed":
|
144 |
+
messages = client.beta.threads.messages.list(thread_id=st.session_state.tech_thread_id)
|
145 |
+
for message in reversed(messages.data):
|
146 |
+
if message.role == "assistant":
|
147 |
+
reply = message.content[0].text.value
|
148 |
+
st.session_state.tech_messages.append({"role": "assistant", "content": reply})
|
149 |
+
try:
|
150 |
+
json_data = json.loads(reply.strip("`json "))
|
151 |
+
st.session_state.tech_results = json_data
|
152 |
+
except:
|
153 |
+
st.session_state.tech_results = []
|
154 |
+
break
|
155 |
+
else:
|
156 |
+
st.error(f"β Technical Assistant failed: {run_status.status}")
|
157 |
+
st.rerun()
|
158 |
+
except Exception as e:
|
159 |
+
st.error(f"β Error: {e}")
|
160 |
+
|
161 |
+
# ------------------ Filters and Display ------------------
|
162 |
+
if st.session_state.tech_results:
|
163 |
+
disciplines = sorted(set(d.get("discipline", "") for d in st.session_state.tech_results))
|
164 |
+
selected_discipline = st.selectbox("π Filter by discipline", ["All"] + disciplines)
|
165 |
+
page_size = 8
|
166 |
+
page_num = st.number_input("Page", min_value=1, step=1, value=1)
|
167 |
+
|
168 |
+
filtered = [r for r in st.session_state.tech_results if selected_discipline == "All" or r.get("discipline") == selected_discipline]
|
169 |
+
paged = filtered[(page_num - 1) * page_size : page_num * page_size]
|
170 |
+
|
171 |
+
st.markdown("---")
|
172 |
+
st.subheader("π Drawing Results")
|
173 |
+
cols = st.columns(4)
|
174 |
+
for i, item in enumerate(paged):
|
175 |
+
with cols[i % 4]:
|
176 |
+
st.markdown(f"**{item['drawing_number']}**")
|
177 |
+
st.markdown(f"_Discipline: {item['discipline']}_")
|
178 |
+
st.caption(item.get("summary", ""))
|
179 |
+
for url in item.get("images", [])[:1]:
|
180 |
+
st.image(url, caption="Thumbnail", use_container_width=True)
|
181 |
+
else:
|
182 |
+
for msg in st.session_state.tech_messages:
|
183 |
+
with st.chat_message(msg["role"]):
|
184 |
+
st.markdown(msg["content"], unsafe_allow_html=True)
|